Medical Image Registration (Master Thesis)

    Image registration is the process of overlaying two or more images of the same scene taken at different times from different viewpoints.Conceptually, it is based on aligning a target image to a source image by determining the transformation that maps points in the target image to points in the source image.


    Algorithms for image registration typically define a measure of similarity between the images and then seek to optimize this measure over the space of valid transformations. A fundamental step in image registration is the selection of a similarity measure. There are many methods for defining the similarity measure depending on the features. Instead of using a similarity measure that is general and can be used in different types of problems, we can learn the similarity measure by using training data. Our method is based on distance metric learning (DML). The objective of distance metric learning is to learn a distance metric that will separate the training data while satisfying some given constraints.

    Full thesis: Thesis.pdf

    Supervisor: Nathan D. Cahill.

    Please feel free to contact me if you have any corrections, questions or suggestions.